Hidden Valley Processes Feb 21, 2007 Erik Brubaker FTK meeting
Aside: FTK training production After last meeting, Francesco managed to turn off unneeded calorimeter sim, digi, reco. Jobs are faster (factor of ~3-4) and output smaller (factor of ??)! UC tier3 cluster had dCache problems for ~1 wk, putting a big dent in production progress. Right now I estimate almost 30M events done—time to ship them off to Pisa and check patterns! February 21, 2007 FTK Meeting
Hidden Valley models Matt Strassler (U Washington) explores “Hidden Valley” model space—regions having light particles, but so far inaccessible experimentally for some reason. One class of HV models has light scalar Higgs decaying into new long-lived (3 ps-3 ns) particles. (hep-ph/0605193) If those new particles decay to jets, the event signature is 2 pairs of jets, each pair sharing a displaced vertex. February 21, 2007 FTK Meeting
HV process features Using mH = 100 GeV, mS = 40 GeV, tS 30 ps. All partons (4/ev) Lowest pT Highest h Very low! February 21, 2007 FTK Meeting
Truth tagging information HV events have nearby vertices at large distance from the beamline… Decay vs Prod Vertex February 21, 2007 FTK Meeting
Event Selection for “offline” analysis Require ≥ 2 jets w/ |h| < 2.5, pT > 15 GeV. Find ≥ 1 pair of jets matched to vertices w/ Dx < 2 cm, avg(r) > 4 cm. Vertices used are B _production_ vertices. Selection is 60% efficient for this benchmark sample. Becomes 96-98% efficient after LVL1 trigger cuts. February 21, 2007 FTK Meeting
Benchmark signal rates Model LVL1 trigger eff. LVL1 rate@10^33 Ev/yr (10^7 s) mH=100; Br(HSS)=1-100%; mS=40; tS=30 ps 4j40: 0.010 [6e-6,6e-4] 60-6000 4j25: 0.033 [2e-5,2e-3] 200-20000 mH=130; Br(HSS)=1-100%; mS=40; tS=30 ps 4j40: 0.017 4j25: 0.057 [2e-5,1e-3] mH=170; Br(HSS)=0.1-1%; mS=40; tS=30 ps 4j40: 0.028 [7e-7,7e-6] 7-70 4j25: 0.104 [2e-6,2e-5] 20-200 No FTK FTK No FTK FTK No FTK FTK LVL1 is the bottleneck, so these numbers tell most of the story February 21, 2007 FTK Meeting
Background to HV process No background processes with the same signature, including shared displaced vertex. BG contributions at some level from KS, L decays in flight + accidentals? Misreconstructed primary vertex? Overlapping jets from different primary vertices? All _should_ be manageable with time and effort, but hard to get an estimate for this study! Existing SHERPA jet samples lack B hadron info in ntuple. Have to regenerate samples—move operation to UC tier3 etc. Make b-enriched. February 21, 2007 FTK Meeting
Next steps Background study for Hidden Valley: Generate b-enriched sample. Check rate to pass cuts using B _decay_ vertex, which is still an overestimate. Perform multi-threshold study a la Kohei. Looks like maybe not needed for HV? LVL1 rates shared with H->hh->4b. February 21, 2007 FTK Meeting
Additional Materials
Different benchmark point Using mH = 130 GeV, mS = 40 GeV, tS 30 ps. Slightly harder than mH = 100 GeV, but not much… February 21, 2007 FTK Meeting
BOLD = I have a 100k-event sample… HV benchmark models Production Process Higgs mass Br(h ==> SS) Mass of S Lifetime of S S decay channels Comments All standard model Higgs production gg ==> h, qq==>qqh, qq==>Wh, Zh 80 GeV; 100 GeV; 130 GeV 100%; 10 %; 1% 20 GeV; 40 GeV 3 ns; 300 ps; 30 ps; 3 ps bb; tau tau An 80 GeV higgs decaying to SS with lifetime less than 300 ps may well be ruled out; LEP has not published limits 170 GeV 1%; 0.1 % 20 GeV; 40 GeV; 65 GeV 3 ns; A Higgs that can decay to WW, ZZ cannot easily have a large Br to SS BOLD = I have a 100k-event sample… February 21, 2007 FTK Meeting
Signal rates Production cross section is gg->H, depends on mH H->SS is 0.1%--1% for high mass, 1%--100% for low mass Higgs Analyze with our full machinery to get LVL1 trigger efficiency for 4j@40GeV (ATLAS default) and 4j@25GeV (with FTK) LVL2 efficiency depends on S lifetime, but e.g. a single tag requirement should be highly efficient Assume 10^33—low LHC luminosity… February 21, 2007 FTK Meeting